agentskills.io Skill Frontmatter: 9 Key Metadata Fields Explained

The agentskills.io skill frontmatter requires nine mandatory YAML fields—name, description, domain, subdomain, tags, version, author, license, and nist_csf—defined between triple-dash delimiters at the top of every SKILL.md file.

In the mukul975/Anthropic-Cybersecurity-Skills repository, each cybersecurity skill is defined in a SKILL.md file that begins with standardized YAML frontmatter. This agentskills.io skill frontmatter block sits between triple-dash delimiters at the very top of every skill definition and provides the canonical metadata necessary for automated cataloging, discovery, and framework mapping on the platform.

Mandatory Metadata Fields for agentskills.io Skill Frontmatter

The frontmatter schema is strictly enforced for proper registration. The following nine fields constitute the core metadata structure parsed by agentskills.io:

name

The name field serves as the unique identifier and URL slug for the skill. It must be unique across the registry and typically uses kebab-case formatting. For example, in skills/triaging-vulnerabilities-with-ssvc-framework/SKILL.md, the value is triaging-vulnerabilities-with-ssvc-framework.

description

This field contains a concise, human-readable summary of the skill's functionality. It should clearly explain what the skill does, such as describing a vulnerability triage methodology using the SSVC framework.

domain and subdomain

The domain field provides high-level classification (e.g., cybersecurity), while subdomain offers granular categorization within that domain (e.g., vulnerability-management). Together, these create a hierarchical taxonomy that enables structured browsing in the agentskills.io interface.

tags

An array of free-form keywords that facilitate search and filtering. Tags include relevant technologies, frameworks, or concepts like ssvc, cisa, cvss, and risk-management. These keywords are critical for cross-referencing skills and enabling filtered views in the registry.

version

Specifies the skill version using semantic versioning or simple string notation, such as '1.0'. This field helps track iterations and ensures users reference the correct implementation of a evolving skill.

author

Identifies the creator using a name or handle (e.g., mahipal). This attribution field supports community contribution tracking and maintains provenance for each skill in the open-source collection.

license

An SPDX-compatible license identifier that governs the skill's usage rights. Standard values include Apache-2.0, ensuring clear legal terms for redistribution and modification across the ecosystem.

nist_csf

Maps the skill to specific NIST Cybersecurity Framework sub-categories, such as ID.RA-01, ID.RA-02, ID.IM-02, and ID.RA-06. This alignment helps organizations identify which framework requirements a skill addresses, supporting compliance-driven workflows.

Parsing agentskills.io Skill Frontmatter Programmatically

Extracting Metadata with Python

The YAML frontmatter can be parsed using standard libraries. The text between the first two --- delimiters contains the metadata dictionary, as implemented in helper functions across the repository:

import yaml
from pathlib import Path

def load_skill_frontmatter(skill_path: Path):
    """Return the YAML frontmatter dict from a skill's SKILL.md."""
    text = skill_path.read_text()
    # Frontmatter is between the first two '---' lines

    frontmatter = text.split('---')[1]
    return yaml.safe_load(frontmatter)

skill_file = Path('skills/triaging-vulnerabilities-with-ssvc-framework/SKILL.md')
metadata = load_skill_frontmatter(skill_file)
print(metadata['name'])          # triaging-vulnerabilities-with-ssvc-framework

print(metadata['tags'])          # ['ssvc', 'vulnerability-triage', ...]

Filtering Skills by Tags via Bash

For quick command-line discovery without Python dependencies, you can grep through the skills directory to find files containing specific tags:

#!/usr/bin/env bash

# Find all skills that include the tag "cisa"

grep -rl '^tags:' . | while read -r f; do
  if grep -q 'cisa' "$f"; then
    echo "$(basename "$(dirname "$f")")"
  fi
done

File Structure and Schema Implementation

In the mukul975/Anthropic-Cybersecurity-Skills repository, every skill directory contains its own SKILL.md file following the identical frontmatter schema. The file skills/triaging-vulnerabilities-with-ssvc-framework/SKILL.md demonstrates a fully populated frontmatter block with all nine mandatory fields defined above. This consistent structure across skills/*/SKILL.md files enables the agentskills.io platform to reliably ingest and render cybersecurity skills without parsing variations.

Summary

• The agentskills.io skill frontmatter requires nine mandatory YAML fields: name, description, domain, subdomain, tags, version, author, license, and nist_csf.
• These fields appear between triple-dash delimiters at the top of every SKILL.md file in the mukul975/Anthropic-Cybersecurity-Skills repository.
• The name field acts as the unique slug, while tags and nist_csf enable advanced filtering and NIST Framework mapping.
• You can programmatically extract this metadata using Python's yaml library to split on --- delimiters, or use standard Unix text processing tools for quick searches.
• All skill definitions follow this canonical schema, located in skills/*/SKILL.md, for automated cataloging on the agentskills.io platform.

Frequently Asked Questions

What is the purpose of the nist_csf field in agentskills.io skill frontmatter?

The nist_csf field maps skills to specific NIST Cybersecurity Framework sub-categories, such as ID.RA-01 or ID.RA-06. This alignment allows organizations to quickly identify which skills address particular framework requirements, facilitating compliance and risk management workflows within the agentskills.io ecosystem.

Can I add custom fields to the agentskills.io skill frontmatter?

While the nine core fields are mandatory for proper registration on agentskills.io, authors may add optional custom fields beyond the required schema. However, according to the repository structure, these additional fields will not be parsed or displayed by the standard platform interface unless explicitly supported by custom integrations.

How does the name field differ from the filename in SKILL.md?

The name field provides a canonical unique identifier used as the URL slug and registry key, whereas the directory name and SKILL.md filename follow filesystem conventions. In the mukul975/Anthropic-Cybersecurity-Skills repository, the name value should match the directory name (e.g., triaging-vulnerabilities-with-ssvc-framework) to ensure consistency in the agentskills.io catalog.

What license identifiers should I use in the license field?

The license field requires SPDX-compatible identifiers, such as Apache-2.0, MIT, or GPL-3.0. Using standard SPDX codes ensures that automated legal compliance tools can correctly interpret the usage rights for each skill in the agentskills.io registry, as demonstrated by the Apache-2.0 value in the SSVC triage skill.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}